The retina is a remarkable piece of neural tissue, providing the gateway through which all visual information enters the mammalian nervous system. It is complex and highly organized, yet far more accessible to characterization than the brain. Furthermore, with its easy-to-manipulate input (images projected onto the retina) and well-defined output (neural code of the optic nerve), the retina lends itself beautifully to studying one of the fundamental questions of neuroscience: how do neurons encode information?
The retina is a multi-layered structure with intricate neural circuitry. In the outer layer are photoreceptors that translate light information into electrical signals, which are passed through a series of interneurons residing in the middle layer, and finally to the retinal ganglion cells (RGCs) in the inner layer. It is the RGCs whose axons form the optic nerve. Rather than simply relaying the raw light intensity values from the photoreceptors, the RGCs actually perform a wide variety of computational transformations to the signal, resulting in over 20 streams of parallel information being sent to the brain. These different types of RGCs can be classified using their distinct morphological features, expression of molecular markers, or electrophysiological functions. There is strong evidence that the morphology and gene expression of RGCs directly relate to their physiological response properties. However, until we have an RGC classification scheme that incorporates these three aspects of neurons, we cannot fully understand how the retina develops, nor can we effectively explore the changes in the retina if it is damaged through disease or mutation.
In this thesis, I make progress in three major avenues that contribute to the understanding of mouse RGCs. One of the key methods I use is the characterization of physiological properties of RGCs using a large-scale multi-electrode array (MEA) recording system. In the first section, I describe my work that improves our ability to functionally characterize mouse RGCs. One project involves the development of a meta-analysis that compares white noise RGC responses across retinas showing we can consistently track 5 cell types across preparations. The second project describes my work developing an analysis pipeline for Direction Selective RGCs and improving the characterization of these well-studied subclasses. In the second section, I develop a technique to link the functional RGC with the morphological RGC on the MEA, using channelrhodopsin stimulation to link anatomical information to the cell’s spiking information. In the third section, I used MEA recordings and my new analyses to determine how specific genes contribute to retinal circuitry. In one case I show a loss of all Direction Selective RGCs in a cell adhesion molecule (DSCAM) knockout mouse model, and in another case I show a loss of only one subtype of Direction Selective RGC in a developmental transcription factor (Satb1/Satb2) double knockout mouse model.